X-Authentication-Warning: delorie.com: mail set sender to geda-user-bounces using -f
X-Recipient: geda-user AT delorie DOT com
Date: Wed, 7 Sep 2016 13:22:15 +0200 (CEST)
From: Roland Lutz <rlutz AT hedmen DOT org>
To: "Vladimir Zhbanov (vzhbanov AT gmail DOT com) [via geda-user AT delorie DOT com]" <geda-user AT delorie DOT com>
Subject: [geda-user] Attribute namespaces (was: Can an attribute be attached to text for
 later inclusion in gnetlist backend?)
In-Reply-To: <20160906213426.GA10224@localhost.localdomain>
Message-ID: <alpine.DEB.2.11.1609071254540.1888@nimbus>
References: <CAGde_xMkqgbbfH81MPOLPfTaui0wRvmUctk31r-eE3=fQ+U0pA AT mail DOT gmail DOT com> <A9C29BBE-B381-4BD1-BD54-E0E27DF1307C AT noqsi DOT com> <20160823053301 DOT 865f671a1b40b5a422e59ce7 AT gmail DOT com> <da433c1d-c711-e0d8-f9ff-a6e843bfe266 AT sbcglobal DOT net>
 <AB0B2DAD-9075-4AEC-B33E-A57DA050B079 AT noqsi DOT com> <CAGde_xOYrkv-4eWyR4OOTT+XQMPcr4MxmT1xomB9uCneZBCT6A AT mail DOT gmail DOT com> <20160824185818 DOT GD14293 AT localhost DOT localdomain> <CAGde_xNfx_VmpWTm6EwHac2QaKQCRefs1cKK=s9gE8OOuMiWdA AT mail DOT gmail DOT com>
 <20160831221409 DOT GA2585 AT localhost DOT localdomain> <CAGde_xMDpUFy5P05Mg+zmzDtvbshvZAghR13F4UkSxxMqtw7yw AT mail DOT gmail DOT com> <20160906213426 DOT GA10224 AT localhost DOT localdomain>
User-Agent: Alpine 2.11 (DEB 23 2013-08-11)
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
Reply-To: geda-user AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: geda-user AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

On Wed, 7 Sep 2016, Vladimir Zhbanov (vzhbanov AT gmail DOT com) [via 
geda-user AT delorie DOT com] wrote:
> Having been thinking before of all the ambiguous attribute names we have 
> in geda where each gnetlist backend has its own vision on their use, now 
> I think every one of them being non-compatible with others should have a 
> different backend-specific name. Say, you cannot use the same toplevel 
> entity/module definition for vhdl, verilog and spice.

I've been thinking about that, too.  Prefixing attributes with the 
"problem domain" they're used for has several advantages: first, it would 
solve the old problem with re-using attributes, like "pinseq" being used 
by both the netlister itself (for slotting) and the SPICE backends, and 
the much-overloaded "device" attribute.  Second, given that there's no 
exhaustive attribute list for gEDA, it would make it more obvious where 
the information in a particular attribute actually goes.

Backends which share a "problem domain" should probably access the same 
attributes (e.g., the 'bom', 'bom2', 'partslist1', 'partslist2', and 
'partslist3' backends), so I suggest using attribute "namespaces" which 
backends can share or not share as appropriate:

     spice:pinseq
     spice:model-name

     bom:device
     bom:nobom

     pcb:footprint

In order to avoid breaking older schematics, backends would obviously have 
to probe for the legacy attribute names, too (though it may be possible to 
hide this in the API).